UFBoot2

UFBoot2 implements an ultrafast bootstrap approximation to provide rapid and robust bootstrap support values for maximum likelihood phylogenetic inference.

Key Features:

• Computational Efficiency: Achieves accelerated bootstrap approximation, reported as 778 times faster than standard bootstrap (SBS) and 8.4 times faster than RAxML rapid bootstrap on tested datasets.
• Improved Accuracy: Refines UFBoot methodology to reduce overestimation of branch support caused by polytomies and severe model violations, producing more reliable and less biased support values than traditional methods.
• Adaptive Resampling Strategies: Incorporates adaptive bootstrap resampling strategies tailored for phylogenomic datasets.
• Convergence Assessment: Employs a stopping rule that assesses convergence of branch support values and determines when sufficient candidate trees have been collected.
• Robustness Against Model Violations: Maintains robust performance under moderate model violations according to simulations, yielding relatively unbiased support values compared to conservative standard bootstrap methods.

Scientific Applications:

• Phylogenomic analysis: Provides fast bootstrap support estimation for large-scale phylogenomic datasets.
• Clade support assessment: Delivers rapid and more interpretable clade support values for maximum likelihood trees, aiding interpretation of branch reliability.
• Evolutionary biology and taxonomy: Supports studies in evolutionary biology, taxonomy, and related fields that require reliable clade support.

Methodology:

Uses the ultrafast bootstrap approximation (UFBoot) with adaptive resampling strategies and a convergence-based stopping rule; performance was evaluated by simulations and benchmarked against standard bootstrap (SBS) and RAxML rapid bootstrap.